PECB Six Sigma Green Belt cheat sheet

Six Sigma performance target: 3.4 defects per million opportunities (DPMO), i.e., a process centered within ±1.5 sigma of specification limits over the long term.

DPMO = (defects / (units × opportunities per unit)) × 1,000,000; use it to compare processes with different complexity.

DMAIC is applied to existing processes; DMADV (Define, Measure, Analyze, Design, Verify) is applied to new designs or redesigns.

Y = f(X1, X2, … Xn): Y is the output/CTQ, X's are the process inputs; improvement acts on X's, control monitors X's.

Lean identifies the eight wastes: defects, overproduction, waiting, non-utilized talent, transportation, inventory, motion, extra processing.

Value-added test: the step changes the form/fit/function of the product, the customer cares about it, and it is done right the first time.

Rolled throughput yield (RTY) = product of each step's first-pass yield; hidden factory losses appear when RTY is far below final yield.

Cost of poor quality (COPQ) includes internal failure, external failure, appraisal, and prevention costs; prevention is cheapest in the long run.

Belt roles: Champion sponsors and removes barriers, Master Black Belt provides technical coaching, Green Belt leads projects part-time, Yellow Belt supports.

A defensible Six Sigma project links to strategic/business goals, has a measurable Y with a baseline, has a clear owner, and is completed in roughly 3–6 months.

Prioritization matrices score candidate projects against weighted criteria (financial impact, customer impact, feasibility, cycle time); higher weighted score wins.

Project charter essentials: problem statement, goal statement, scope/boundaries, business case, team, milestones, and constraints.

Problem statement rules: describe the gap with data, state time, location, and magnitude, and avoid naming a cause or a solution.

SMART goal: specific, measurable, achievable, relevant, time-bound; include the metric, baseline, target, and date.

SIPOC maps Suppliers, Inputs, Process, Outputs, Customers at a high level and defines project boundaries (start and stop points).

Scope creep control: state explicitly what is in and out of scope, and require a charter revision for changes.

Voice of the Customer (VOC) is translated into CTQ characteristics via Kano analysis or CTQ trees; CTQs must be measurable and actionable.

Stakeholder analysis maps influence vs. interest; high-influence/high-interest stakeholders get active engagement and early updates.

Measurement system analysis: repeatability = equipment variation (same appraiser), reproducibility = appraiser variation (between operators).

Gage R&R acceptance: %GRR < 10% acceptable, 10–30% conditionally acceptable depending on application and cost, >30% unacceptable.

Attribute agreement: use kappa or percent agreement; kappa > 0.75 indicates good agreement beyond chance.

Data types: continuous (variable) data supports more powerful tests; discrete (attribute) data needs larger samples.

Sampling: random sampling reduces bias; rational subgrouping captures only common-cause variation within a subgroup.

Baseline capability: Cp = (USL − LSL) / 6σ (spread); Cpk = min[(USL − μ)/3σ, (μ − LSL)/3σ] (centering plus spread).

Process capability requires a stable process in statistical control; capability indices are meaningless for an unstable process.

MSA should be done before capability studies; measurement variation must be small relative to the process and tolerance variation.

Data collection plan specifies what to measure, operational definitions, who collects, when, how many, and how it is recorded.

Root cause tools: 5 Whys, fishbone (Ishikawa) diagram with 6M categories, Pareto chart, and process mapping.

Pareto principle: roughly 80% of problems come from 20% of causes; prioritize the vital few categories.

Hypothesis test logic: H0 assumes no difference/effect; p-value < α rejects H0; α is the risk of a Type I error (false alarm).

Type II error (β) is failing to detect a real effect; power = 1 − β, commonly targeted at 0.80 or higher.

t-test compares means (1-sample vs. target, 2-sample for two groups, paired for before/after on the same units); ANOVA compares 3+ group means.

Chi-square tests association between two categorical variables; use it for attribute data contingency tables.

Correlation coefficient r ranges −1 to +1 and measures linear association only; correlation does not establish causation.

Regression models the relationship Y = b0 + b1X + error; check R², residual plots, and p-values of coefficients.

Control charts separate common from special cause variation; a point beyond control limits or a non-random run signals special cause.

Solution selection uses weighted criteria (impact, cost, risk, time to implement) and a pilot before full deployment.

Design of Experiments (DOE): screening designs (fractional factorial) identify significant factors; full factorial estimates main effects and interactions.

FMEA: RPN = Severity × Occurrence × Detection; higher RPN gets priority action, and severity alone can force action regardless of RPN.

Mistake-proofing (poka-yoke) prevents errors at the source; it is preferred over inspection and training alone.

Kaizen events are short, focused improvement bursts; 5S (sort, set in order, shine, standardize, sustain) organizes the workplace.

Control plan documents the monitored CTQ, specification, measurement method, sample size/frequency, control method, and reaction plan.

SPC control chart choice: X-bar/R for continuous subgroups (n ≤ 8), X-bar/S for larger subgroups, p/np for defectives, c/u for defects.

Standard work and visual management sustain gains; document the new procedure and train operators before handover.

Project closure requires verified results against the goal, control plan ownership transfer, documented lessons learned, and financial validation.